Drastic erosion in functional attributes of tree assemblages in Atlantic forest fragments of northeastern Brazil

Abstract

The long-term dynamics of plant communities remain poorly understood in isolated tropical forest fragments. Here we test the hypothesis that tropical tree assemblages in both small forest fragments and along forest edges of very large fragments are functionally much more similar to stands of secondary growth (5–65-yr old) than to core primary forest patches. The study was carried out in a severely fragmented landscape of the Brazilian Atlantic forest. Nine functional attributes of tree assemblages were quantified by sampling all trees (DBH ⩾ 10 cm) within 75 plots of 0.1 ha distributed in four forest habitats: small forest fragments (3.4–79.6 ha), forest edges, second-growth patches, and primary forest interior areas within a large forest fragment (3500 ha). These habitats were markedly different in terms of tree species richness, and in the proportion of pioneer, large-seeded, and emergent species. Age of second-growth stands explained between 31.4% and 88.2% of the variation in the functional attributes of tree assemblages in this habitat. As expected, most traits associated with forest edges and small forest fragments fell within the range shown by early (<25-yr old) and intermediate-aged secondary forest stands (25–45-yr old). In contrast to habitat type, tree assemblage attributes were not affected by vegetation type, soil type and the spatial location of plots. An ordination analysis documented a striking floristic drift in edge-affected habitats. Our results suggest that conservation policy guidelines will fail to protect aging, hyper-fragmented landscapes from drastic impoverishment if the remaining forest patches are heavily dominated by edge habitat.

Introduction

Habitat loss and fragmentation can drastically alter the structure and composition of tropical tree assemblages (Laurance, 2001). Forest edges and even central portions of small forest fragments are expected to exhibit species-poor tree communities and may be impoverished in terms of emergent (Laurance et al., 2000), understorey (Tabarelli et al., 1999), and hardwood or old-growth tree species (Michalski et al., 2007). Large-seeded trees dispersed by vertebrates, particularly medium to large-bodied frugivores, appear to be vulnerable to population collapse (Cordeiro and Howe, 2001, Melo et al., 2006, Terborgh and Nuñez-Iturri, 2006). Other tree functional groups may also be poorly represented in edge-affected habitats, including species sensitive to habitat desiccation (Ferreira and Laurance, 1997), fire disturbance (Cochrane and Laurance, 2002) and competition with vines and ruderal species (Scariot, 2001, D’Angelo et al., 2004). Consequently, severely fragmented landscapes reduced to archipelagos of small forest fragments are likely to retain but a small, non-random subset of species from the original source flora (Laurance, 2001, Tabarelli and Gascon, 2005). However, the extent to which habitat fragmentation depletes plant populations and truncates the composition of plant assemblages remains poorly understood based on either natural or controlled experiments (see Hobbs and Yates, 2003, Aguilar et al., 2006, Wirth et al., 2007).

One of the most conspicuous features of tree assemblages along forest edges is the elevated abundance and richness of fast-growing pioneer species. Pioneer species account for only 2–3% of tree species richness in continuous forest plots in central Amazonia, but represent over a quarter of the species richness and three times the abundance in forest fragments 1–100 ha in size after 13–17 years of isolation (Laurance et al., 2006a). Some pioneers rapidly increase their abundance in forest edges by over 1000%, although others either remain constant or decline in abundance (Laurance et al., 2006a, Laurance et al., 2006b). Pioneer-dominated tree assemblages or over proliferation of pioneers have also been documented in stable Atlantic forest edges in fragments that are at least 60-yr old (Tabarelli et al., 1999, Oliveira et al., 2004, Grillo, 2005), suggesting that such assemblages may be approaching near-equilibrium conditions, rather than represent transient successional stages.

In edge-dominated habitats, proliferation of pioneer species occur in parallel with phytomass collapse (Laurance et al., 1997), suggesting that assemblages comprised primarily of old-growth, high wood-density trees have been replaced by second-growth or regenerating vegetation (Gascon et al., 2000). Indeed, elevated abundance of pioneer species is a typical feature of early secondary forests younger than 25–30 yrs following slash-and-burn agriculture (sensu Uhl, 1987) or other disturbance events such as large treefall gaps (Brown and Lugo, 1990, Tabarelli and Mantovani, 2000, Howorth and Pendry, 2006). Rates of forest regeneration and other assemblage-level properties of young secondary forests are largely affected by environmental constraints on tree species establishment such as diaspore availability (Nepstad et al., 1996, Vieira et al., 2006), seed and seedling predation (Uhl, 1987), soil infertility (Ganade and Brown, 2002), and competition with grasses and ruderal species (Hooper et al., 2005). Early secondary forests are therefore not homogeneous at any spatial scale (Richards, 1996), but provide suitable habitat for only a biased and impoverished subset of tree species, particularly small-seeded pioneers (Aide et al., 2000, Guariguata and Ostertag, 2001, Tabarelli and Peres, 2002, Martínez-Garza et al., 2005). Furthermore, because secondary forests are dominated by pioneer species sharing a similar set of traits regardless of their taxonomic affinities (Whitmore, 1991, Richards, 1996, Turner, 2001), these assemblages are expected to sustain only a small suite of tree life-history traits found in old-growth forests.

The Atlantic forest of northeast Brazil is now reduced to less than 6% of its original pre-Colombian extent with high levels of forest fragmentation going back to the 16th century (Galindo-Leal and Câmara, 2003). Habitat loss and fragmentation are so striking that 48% of the remaining fragments in this region are <10 ha and very few are >1000 ha (Ranta et al., 1998). Old primary forest fragments surrounded by a homogeneous matrix of sugar-cane plantations offer an excellent opportunity to examine the long-term effects of habitat fragmentation on the functional composition of stabilizing tree assemblages. Here we ask whether habitat fragmentation in an Atlantic forest landscape can effectively promote functional convergence of tree assemblages between forest edges, old-growth forest and early second-growth stands following a known history of slash-and-burn agriculture. In particular, we test the hypothesis that tree assemblages along forest edges and in small forest fragments are much more similar to those of secondary forest stands (5–65-yr old) than to those of forest interior conditions in an exceptionally large fragment. First, we compare nine structural and functional attributes of tree assemblages in these four forest habitats, including species richness and the relative species contribution to different ecological groups of regeneration strategy, vertical stratification, seed dispersal mode, and seed size. Second, we document levels of taxonomic similarity between habitats based on ordination analysis, and identify indicator tree species. Finally, we discuss the processes driving changes in tree assemblages within aging forest fragments, and their implications for biodiversity persistence in fragmented landscapes.